Order Workflows

Wholesale buyers rarely shop item by item. They restock from spreadsheets, scan barcodes during stocktakes, or type SKUs they already know. A single ordering path forces them into the wrong workflow and slows everything down.

Quick Order routes three input methods through one experience: manual entry for known reorders, CSV upload for existing reorder sheets and POS exports, and barcode scanning with separate modes for one-off restocks at the shelf and full-aisle stocktakes. SKUs, UPCs, and EANs are accepted interchangeably, with live validation against the catalog and automatic pack-size rounding.

Gift cards are a product the customer buys and a payment method the recipient uses. The system has to handle both sides cleanly: scheduled delivery, fraud screening before the card reaches the recipient, and partial redemption combined with any other payment method at checkout.

A unified workflow handles the full lifecycle. The buyer chooses an amount, adds a personal message, and schedules the exact send date. Each card is generated, queued, and screened for fraud before being emailed to the recipient. At checkout, recipients redeem the full balance or apply it as a partial payment, with the remainder routed to any other payment method on file.

Every 'place order' click triggers a coordinated sequence across multiple systems. Product data, cart state, address validation, delivery windows, payment processing, fraud screening, ERP order creation, and analytics dispatch all have to happen in real time, in the right order, without the customer feeling any of it.

The architecture coordinates inRiver PIM, Optimizely CMS and Commerce, Dynamics AX, a fraud platform, and ten payment providers through a CQRS pattern with MediatR. The cart is re-validated against the ERP before order creation, the sales order is created in Dynamics AX as the system of record, and the purchase event is dispatched to analytics and CDP both server-side and through the tag manager.

Most commerce platforms try to own the cart and checkout logic, with the ERP as a downstream system that receives the order. When the ERP has decades of business logic already encoded (pricing, discounts, tax, coupons, club cards, gift cards, shipping rates), duplicating that logic in commerce creates two sources of truth and constant drift.

Optimizely Commerce stores cart, inventory, and order data without running any business logic. Every transactional decision is delegated to the legacy ERP through 9 SOAP endpoints: cart price recalculation on every line item change, real-time shipping rate lookup, full tax calculation, coupon validation, club card and gift card balance checks, and order submission with full transaction context.

Pure external fraud detection treats fraud as a generic problem solvable by AI signals alone. Pure internal rules treat fraud as a business problem solvable by static thresholds. Real B2B and specialty commerce needs both: global fraud intelligence to catch sophisticated patterns, and configurable business rules to catch program-specific abuse that external systems cannot see.

Every order runs through two evaluation layers. Forter provides external fraud intelligence using card validation, device and identity signals, and behavioral analysis. The internal rules engine applies configurable thresholds, email domains, brand volume, order frequency, product velocity, and the decision layer combines external validation with internal rules to pass or hold orders, with full audit trails for the operations team.

Regulated commerce can't treat compliance as a single check at checkout. License-required transfers, hazmat, age-restricted, state-restricted, and internationally controlled products each carry their own rules, and those rules interact. A single boolean 'compliant: yes/no' isn't enough.

Every order runs through eleven sequential validation stages, from purchase validation and address correction through ShipCompliant evaluation, age verification, hazmat and FFL fee calculation with exemption logic, and Avalara tax, only releasing to processing after clearing every stage with full audit logs preserved for every check.

High-value B2B customers often need agent assistance during ordering, but most platforms force a binary choice: either the customer completes the order themselves or the agent takes over completely. Neither model fits VIP customer service, where the customer sometimes wants to self-serve and sometimes wants white-glove help.

VIP customers can complete their own checkout or a customer service agent can use an impersonation tool to authenticate as the VIP customer, complete checkout on their behalf, then explicitly unlock the cart, delete saved credit cards, and log out to return the session to a clean state. Both paths route through the same OMS and Stores App for fulfillment.